> I actually thought something like the CLOCK_MONOTONIC or CLOCK_BOOTTIME were
> used for providing uptime to the user, but those do not allow date changes.
You are assuming a certain implementation, and then thinking what properties
the program "must" have. Try thinking the other way around: What properties
the program should have, and then filter out the implementations that don't
satisfy these properties.
> I do not understand the relationship between date and boottime. Might make
> sense for VM's, but for non-VM's I do not get it.
We did the changes
- for year-2038 safety on Linux/{x86,arm},
- so that it is stable across 'sudo date MMDDhhmm' invocations [2],
- so that it doesn't display a fantasy time in VMs [1].
Even if it were only for VMs. VMs are the basis of cloud environments,
and are thus pretty important nowadays.
Bruno
[1] https://lists.gnu.org/archive/html/bug-gnulib/2023-08/msg00064.html
[2] https://lists.gnu.org/archive/html/bug-gnulib/2023-08/msg00067.html
